Optical textures classification of coke microscopic image based on SVM

In the view of characteristics of coke optical texture in micrograph, a classification method, which is based on Support Vector Machine and combining color and texture features, is proposed. Firstly, color features of the coke microscopic images of different optical textures are analyzed. With color features, isotropic and anisotropic components are classified. Then the gray level co-occurrence matrix of anisotropic components is calculated, the texture features (such as entropy) of each anisotropic components are computed. With texture features, each subclass in anisotropic component is further classified. Experimental results show that with the proposed method the classification among different optical textures of coke is more reasonable and effective than traditional techniques, including neural networks.